Journal of Thermal Biology 62 (2016) 76–83
Contents lists available at ScienceDirect
Journal of Thermal Biology
journal homepage: www.elsevier.com/locate/jtherbio
crossmark
Self-heating by large insect larvae?
Nikita L. Cooley, Douglas J. Emlen, H. Arthur Woods
⁎
Division of Biological Sciences, 32 Campus Drive HS104, University of Montana, Missoula, MT, 59812, USA
A R T I C L E I N F O
A BS T RAC T
Keywords:
Heat balance
Body size
Metabolic heat
Critical thermal maxima
Behavioral thermoregulation
Gigantism
Soil
Do insect larvae ever self-heat significantly from their own metabolic activity and, if so, under what sets of
environmental temperatures and across what ranges of body size? We examine these questions using larvae of
the Japanese rhinoceros beetle (Trypoxylus dichotomus), chosen for their large size ( > 20 g), simple body plan,
and underground lifestyle. Using CO2 respirometry, we measured larval metabolic rates then converted
measured rates of gas exchange into rates of heat production and developed a mathematical model to predict
how much steady state body temperatures of underground insects would increase above ambient depending on
body size. Collectively, our results suggest that large, extant larvae (20–30 g body mass) can self-heat by at most
2 °C, and under many common conditions (shallow depths, moister soils) would self-heat by less than 1 °C. By
extending the model to even larger (hypothetical) body sizes, we show that underground insects with masses >
1 kg could heat, in warm, dry soils, by 1.5–6 °C or more. Additional experiments showed that larval critical
thermal maxima (CTmax) were in excess of 43.5 °C and that larvae could behaviorally thermoregulate on a
thermal gradient bar. Together, these results suggest that large larvae living underground likely regulate their
temperatures primarily using behavior; self-heating by metabolism likely contributes little to their heat budgets,
at least in most common soil conditions.
1. Introduction
Like many ectotherms, insects face the problem of attaining
reasonable body temperatures—temperatures that permit adequate
performance, including feeding, growth, and reproduction—in environmental conditions that vary in space and time. In cool environments,
insects may live far down on the left sides of their thermal performance
curves (Deutsch et al., 2008). For insects in such a situation, there are
exponential increases in performance to be gained by finding ways to
achieve higher body temperatures. Insects often do so using behavioral
thermoregulation (May, 1979; Heinrich, 1993), which involves moving
through locally available mosaics of microsites and choosing sets of
conditions that give higher body temperatures (Bakken, 1992). For
individual insects, this is equivalent to choosing sites that increase
inputs, and decrease outputs, to its heat budget—for example, by
choosing microsites with relatively high air temperatures, high levels of
incoming visible and infrared radiation, and relatively warm nearby
objects (Woods et al., 2015) or that give low rates of heat loss by
evaporation and convection (Gates, 1980). In hot environments, by
contrast, the problem is to avoid body temperatures so high that the
insect's performance is declining. In such environments, the insect
needs to decrease inputs of heat and, if possible, increase outputs. In
general, actively seeking out cooler conditions also involves behavioral
⁎
thermoregulation.
Insects may also warm themselves up significantly using heat
produced by their own metabolism. Whether they do so depends
primarily on mass-specific intensity of metabolism, body size, and
rates of heat loss, which are lower across well-insulated surfaces. Birds
and mammals, most of which have high mass-specific metabolic rates
and large body sizes (compared to insects), are the consummate
endothermic homeotherms. Most produce enough heat to sustain
constant body temperatures in the range of 35–40 °C. Insects of course
are smaller and, with some significant exceptions (see below), have
mass-specific metabolic rates that are an order of magnitude or more
lower per unit body mass than those of birds and mammals (Robinson
et al., 1983); in general, their body temperatures are much more closely
coupled to ambient environmental conditions. Much of the time, they
are ectothermic poikilotherms. Even so, it may be possible for insects to
produce significant metabolic heat. The best examples are large flying
adults, which can have extraordinarily high mass-specific metabolic
rates (Bartholomew and Casey, 1978). Adults of several species of
moths maintain thoracic temperatures above 38 °C when flying, which
is possible because their thoracic muscles produce heat at such high
rates (Heinrich, 1970; Casey, 1981). Indeed, some flying insects
produce so much metabolic heat that they risk overheating (Heinrich,
1980, 1993); the difficulty then is to shed heat rapidly enough. This
Corresponding author.
E-mail address: [email protected] (H.A. Woods).
http://dx.doi.org/10.1016/j.jtherbio.2016.10.002
Received 26 April 2016; Received in revised form 8 October 2016; Accepted 16 October 2016
Available online 17 October 2016
0306-4565/ © 2016 Elsevier Ltd. All rights reserved.
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
may explain why some large beetles and moths fly only at night.
Because holometabolous insect larvae do not power locomotion
nearly as intensely as adults, their mass-specific metabolic rates while
active generally are much lower. For example, a fifth-instar caterpillar
of Manduca sexta consumes 30–50 µmol O2 g−1 h−1 (Greenlee and
Harrison, 2005) whereas an adult M. sexta in free flight consumes
about 2 mmol g−1 h−1 (Heinrich, 1971a), which is 40 – 65× greater on
a mass-specific basis. Nevertheless, several factors may allow larvae to
self-heat. First, larvae generally have simple sac-like bodies, with
relatively small ratios of surface area to volume and therefore relatively
small surfaces across which to lose heat. Second, not all larvae have low
mass-specific metabolic rates; for example, some herbivorous caterpillars feed and digest at very high rates, with that activity supported by
high mass-specific rates of metabolism (Kingsolver and Woods, 1997).
Third, larvae of some species can grow to very large absolute body sizes
(always larger than the adults into which they metamorphose). Large
insects generally have higher absolute metabolic rates (Chown et al.,
2007) and thus produce more metabolic heat, possibly causing them to
self-heat to greater extents. Fourth, larvae of many species live in
confined spaces—underground, in rotting vegetation or wood, in
constructed shelters—where there is little potential for evaporative or
convective cooling (for example of bird eggs in such an environment,
see Seymour and Bradford, 1992). In effect, those confined spaces can
act as insulation.
We examine the potential for larval self-heating using the Japanese
rhinoceros beetle (Trypoxylus dichotomus), which occurs in Japan,
Taiwan, Korea and eastern China. On Honshu island, Japan, where
these animals have been best studied, adult males defend wounds on
the sides of Fraxinus sp. and other trees, which serve as feeding sites
(Hongo, 2007a, 2007b; McCullough, 2012). Females visit these territories to feed and mate with males, leaving later to lay eggs in the soil
near decaying humus (Karino et al., 2004). Eggs are laid between July
and September, and larvae feed below ground until June – July of the
following year (Plaistow et al., 2005). As they feed, larvae orient
towards high concentrations of CO2 (Kojima, 2015), which appears
to lead them towards high quality, or especially fermented, humus.
This also can lead to large aggregations of larvae (Kojima et al., 2012,
2014; Kojima, 2015), at depths down to 25 cm in the soil. At the end of
their final (3rd) larval instar, animals construct brittle pupal cells,
which are vulnerable to damage by other burrowing larvae.
Because of their extreme size, and the fact that they develop
underground in decomposing—hot—organic matter, Trypoxylus larvae
may be prone to overheating. First, larvae can reach over 30 g (D.
Emlen, personal observation). If any larvae self-heat as a result of high
metabolic rates and large body size, these are likely candidates. Second,
their soil microhabitats provide little potential for evaporative or
convective cooling and may insulate larvae from rapid heat loss by
conduction. Using flow-through and thermolimit respirometry
(Lighton and Turner, 2004), we measured larval metabolic rates and
values of CTmax. In addition, we examined the ability of larvae to
behaviorally thermoregulate on a thermal gradient bar. Lastly, we
developed and analyzed a mathematical model of larval heat balance
that predicts increases in body temperature from self-heating. We used
the model to evaluate self-heating in soils of different temperatures and
across a set of realistic body sizes. Finally, we extended the model to
ask how much very large insect larvae (larger than extant) would selfheat.
from a 1:1 mixture of organic hardwood compost (Hiroki Gotoh,
personal communication) and quick-fermented hardwood sawdust
(Emlen et al., 2012). Additionally, eggs were collected from a local
laboratory colony and allowed to grow to the third instar. 110 larvae
were collected and placed in plastic jars (1 L) containing the previously
described substrate mixture. We kept approximately 50 larvae at room
temperature (22 °C) and another 60 in a temperature-controlled
growth chamber at 10 °C to postpone pupation. Larvae were pulled
from the growth chamber as needed and held at room temperature for
24 h before being used in experiments.
2.2. Larval metabolic rates
We estimated larval metabolic rates from rates of carbon dioxide
emission, using flow-through respirometry. CO2 levels were measured
by an infrared gas analyzer (LI-7000, LI-COR, Lincoln, Nebraska) set
up in differential mode. In this mode, dry, CO2-free air from a cylinder
of compressed breathing air (Norco, Boise, ID, USA) was first passed
through the instrument's reference side, then past the larva and
returned to the instrument's measurement side. The gas analyzer was
calibrated using pure N2 and 2000 parts per million (p.p.m.) CO2 in N2
(NorLab, Boise, ID, USA). Flow rates of gas were 200 mL min−1 (STP)
and were regulated by a mass flow controller (Unit UFC-1100,
500 mL min−1 maximum flow rate, Yorba Linda, CA), which was
controlled by a separate set of electronics (MFC-4, Sable Systems).
The flow controller was recently calibrated at the factory, and we
checked its flow at 200 mL min−1 against a bubble flow meter. Analog
signals from the LI-7000 were sent to an A/D converter (UI2, Sable
Systems) and then recorded using the ExpeData software (Sable
Systems). Rates of CO2 emission were converted to L min−1 O2
consumption assuming a respiratory exchange ratio of 0.70 (Chown
et al., 2007) and then transforming to μW of heat output using the
conversion of 19.7 kJ L−1 of O2 consumed (Schmidt-Nielsen, 1997)
×60 s min−1×109 μJ kJ−1. Errors in estimates of the heat output per
unit O2 consumed will lead to an error of at most 7% in rate of heat
production.
Carbon dioxide emission was measured for ten rhinoceros beetle
larvae at room temperature (22 °C). Each larva was taken from its
rearing jar (no prior period of food deprivation) and placed into a
horizontal 110-mL cylindrical glass chamber (length 160 cm, diameter
3 cm) sealed by a Teflon end cap with two built-in O–rings. Air entered
through a hole in the end cap and exited at the other end through a
drawn out taper in the glass that was connected to Bev-a-line tubing, so
that the chamber was flushed approximately twice per minute. Larvae
were weighed and placed in the chamber 5–10 min prior to measurement to allow them to settle down. Sexes were unknown. Each larva's
CO2 output was measured for 15 min, during which time it was free to
move. Movement had only slight effects on CO2 emission, probably
because movements themselves are very slow. CO2 measurements were
reported as the average p.p.m. for the last 7 min of each experiment.
Because the larvae were not post-absorptive and we included periods
during which they were moving, the metabolic rates derived from these
measurements are neither resting nor standard; rather, they reflect
digestion, absorption, and moderate activity. These conditions are the
most appropriate for estimating rates of heat output during normal
activity in the field.
2.3. Critical thermal maxima
2. Materials and methods
The CTmax was determined for six larvae using thermal ramping.
Because these beetle larvae are some of the largest insects ever
measured, some in excess of 25 g, we developed a protocol based on
work by Terblanche et al. (2007) and using the thermolimit analysis
proposed by Lighton and Turner (2004) (see below). The primary
considerations were to ramp slowly enough that larval body temperatures were in thermal steady-state with the temperature of the air in the
2.1. Animals
Larvae of the Japanese rhinoceros beetle were purchased from a
commercial insect distributor (Yasaka Kabuto Kuwagata World,
Hamada City, Japan) and reared to adulthood in the laboratory.
Individuals were placed in plastic jars (1 L) containing substrate made
77
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
the absolute difference sums (ADS) of both the CO2 emission and
activity data for each larva (Lighton and Turner, 2004).
2.4. Behavioral thermoregulation experiment
In order to test whether larvae adjusted their body temperature by
moving to warmer or cooler locations, we exposed them to composted
substrate distributed on a custom-built thermal gradient bar. The bar
was constructed from a 20-kg block of aluminum of dimensions
0.914 m
(length)×0.305 m
(width)×0.0254 m
(thickness).
Temperatures at the two ends of the bar were fixed by circulating
temperature-controlled water through 13 mm diameter threaded holes
drilled completely through each end across the width. Water temperature was regulated using a programmable Polystat® temperature
controlled recirculating bath (Cole-Parmer) and a digital refrigerated
recirculating chiller (VWR Scientific). The temperature on one end was
set to 5 or 10 °C, while the other end was set to 50 °C. The aluminum
block was insulated on the bottom and sides by pieces of Styrofoam.
The surface was divided into four lanes using 2.54 cm tall Plexiglas
dividers to restrict each larva to lengthwise movement and keep the
larvae from interacting. Lanes were filled with 2–3 cm depth of a dry
1:1 mixture of organic hardwood compost and quick-fermented hardwood sawdust substrate that was allowed to equilibrate to the linear
gradient temperature. While they were moving in the substrate, larvae
were in direct contact with the aluminum.
To ensure that we had created a linear thermal gradient in only the
lengthwise direction, 16 type T thermocouples were inserted into 3 mm
diameter holes drilled into the underside of the aluminum block. The
thermocouple holes were distributed along the middle of the aluminum
block every 11.43 cm, as well as across the block (in two places) every
6.1 cm. To ensure good thermal contact, holes were first filled with
thermal paste. The steepness of the gradient was 0.4 °C (range 0.39–
0.44) per cm. A total of 35 larvae was used to determine temperature
preference along this thermal gradient, while 28 larvae were used in
control runs (constant temperature of 22−22.5 °C everywhere in the
arena).
Individual larvae were introduced into the gradient bar facing a
random direction. Runs were video-recorded from above using a
webcam. The larvae displaced substrate as they traveled along their
lane, which allowed us to observe their movements from the camera.
Recordings lasted either 1 or 2 h, and still images were recorded every
10 s. These images were analyzed in ImageJ using a manual tracking
plugin, allowing us to track the paths of each individual larva. We then
converted each position to an environmental temperature.
Fig. 1. Schematic of apparatus for larval ramping experiments. Larvae were ramped in a
water-jacketed flask whose temperature was set by a programmable temperaturecontrolled recirculating water bath. Carbon dioxide emission was measured using a
flow-through system, temperature was measured by a type-T thermocouple, and activity
was monitored by an infrared motion detector.
flask, but quickly enough to reduce the effects of accumulated exposure
to high but sublethal temperatures and to the experimental apparatus
(Rezende et al., 2011). To determine a good ramp rate, test ramps were
performed in which a larva was placed in the chamber with one
thermocouple taped to the cuticle and another measuring the surrounding air temperature in the chamber. We decided to use a ramp
rate of 0.08 °C min−1, which was fast enough that individual ramps
could be done in approximately 6 h but slow enough that larvae were
largely in thermal steady-state with the flask air. However, even at this
slow ramping rate, body temperatures of the larvae lagged 1–1.5 °C
behind air temperature. It is unclear whether the temperature lag
reflected larval thermal inertia or evaporative cooling from the cuticle.
In general, rates of transcuticular evapotranspiration in insects are
small, although they could be higher in soil-dwelling stages, like larval
Trypoxylus, that are normally exposed to high ambient relative
humidities. The temperature ramp began at 20 °C and ended at
50 °C, giving a total ramp time of 6 h and 25 min
Larvae were ramped individually in a 250-mL, water-jacketed, glass
Erlenmeyer chamber (see Fig. 1). The chamber was attached to a
programmable Polystat® temperature-controlled recirculating water
bath (Cole-Parmer). During each ramp, we measured rates of CO2
emission (using the system described above), actual air temperature,
and levels of activity by the larva. Flow rates of dry air were
350 mL min−1, with air directed in through a metal cannula in the
rubber stopper and then through a plastic tube to the bottom of the
flask; outgoing air went out through another metal cannula that just
pierced the stopper. Temperature was measured by a type T thermocouple inserted into the flask's stopper through a small hole and
attached to a thermocouple meter (TC-1000, Sable Systems). Levels of
activity were monitored using an infrared activity detector (AD-2, Sable
Systems) modified so that the emitter and detector both were mounted
on long wires. The wires were glued into the flask stopper so that the
emitter and detector both were positioned about 1 cm above the larva.
Activity was monitored in 4 of the 6 individual ramps. To identify
CTmax, we used Lighton and Turner's (2004) approach. As an insect's
temperature rises, it eventually undergoes rapid physiological change
(identified as CTmax) during which (simultaneously) rates of CO2
emission fall sharply, gross motor movements cease, and control over
the spiracles is lost (CO2 trace suddenly stops showing high-frequency
fluctuations). To aid in identifying these breakpoints, we also calculated
2.5. Analytical model of larval heat balance
A key aspect of our study is to ask about the thermal biology of
larvae in the size range we studied (up to approximately 30 g) and of
larvae larger than those that we measured directly. We approached this
problem by extrapolating from our measured values—by scaling them
up to hypothetical large larvae. The two specific questions were: (1) Do
larvae of T. dichotomus produce enough metabolic heat to warm
themselves significantly underground? (2) More generally, how big
must underground insects evolve to be before heat from their own
metabolism raises their temperatures significantly?
To analyze these questions, we developed a model of larval heat
balance. Larvae are modeled as metabolizing cylinders of length L and
diameter D buried at depth z in the soil (Fig. 2). For an animal in soil,
the heat generated is equal to the heat lost by conduction to the soil and
the heat stored in its tissues:
Qgen=Qcond +Qst .
(1)
The core-skin temperature gradient for a cylindrical object with
distributed heat generation (Bird et al., 2002) is
78
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
stable and metabolic heat generation is constant. We found this steadystate body temperature using the ode function from the deSolve
package (Soetaert et al., 2010) in R. The temperature-dependence of
Qgen was modeled by introducing a temperature factor describing the
factorial increase in metabolic rate as body temperature rises above
20 °C (see Supplemental Fig. 1). The relationship between metabolic
rate and body temperature was a fitted sigmoid derived from the mean
values of metabolic rate calculated from the ramping experiments used
to establish CTmax.
2.5.2. Assumptions
The model above makes seven key assumptions. (1) The beetle larva
is cylindrical and is in direct contact with the soil surrounding it. In
fact, beetles often curl into more compact masses. Larvae do not
excavate large air-filled chambers; usually a significant proportion of
the larval cuticle is in contact with the surrounding soil. (2) Beetle
larvae stay in the same location in the soil until they reach steady-state
body temperatures, and the soil temperature profile itself is unchanging. Because Eq. (9) describes transient changes in body temperature,
these assumptions could be relaxed in a more elaborate analysis. (3)
Levels of O2 and CO2 in the soil do not alter or limit rates of
metabolism by larvae. For most larvae most of the time, this will be
true. Gas levels in soils generally match the composition of local air
aboveground, as long as the soil is relatively dry. In wet soil, high rates
of soil respiration together with low rates of gas transport can give low
levels of O2 and high levels of CO2 (Campbell and Phene, 1977;
Greenway et al., 2006). In large piles of rapidly fermenting soils
(compost), levels of CO2 can reach 2 – 4% and O2 levels are drawn
down by the same amount (Macgregor et al., 1981). For resting insects,
the critical partial pressure of O2 (below which metabolism is
depressed) is usually quite low—on the order of 3–5 kPa O2
(Harrison et al., 2014)—suggesting that low O2 generally will not be
a problem. High levels of CO2 may cause greater physiological
disturbance, by anaesthetizing larvae and leading to long-term metabolic changes (Colinet and Renault, 2012). (4) All heat transfer in the
soil is by conduction, not by the evaporation, transport, and condensation of water vapor, which has been shown (Westcot and Wierenga,
1974) to account for a significant fraction of heat flux in soil. The errors
introduced by this assumption likely are small, because the methods
used to determine soil conductivity (see Seymour and Bradford, 1992;
Ahn et al., 2009) include the effects of water vapor fluxes; in other
words, the effects of water vapor transport, while not explicitly included
in our model, are captured in the range of values of kS we used. (5) Heat
transport inside larvae is by conduction, not circulation of the
hemolymph. Clearly some heat will be moved by hemolymph, and this
could be modeled using higher effective kL ( > 0.5 W C−1 m−1). (6)
Larval metabolic rate depends on Tc , not on the temperature gradient
between Tc and Ts . (7) Larval density and aspect ratio are invariant
across body sizes.
Fig. 2. Schematic of a larva in soil. The larva is idealized as a cylinder of length L and
diameter D positioned at depth z below the surface of the soil. The larva has surface
temperature Ts and the soil surface has temperature Ta .
Tc−Ts=
qR2
4kL
(2)
where q is the heat generated per unit volume, q = Qgen / V and kL is the
conductivity of larval flesh. Multiplying both sides by V and rearranging gives
Qgen=
4kL V (Tc − Ts )
.
R2
(3)
Conduction from the cylinder is described by
(4)
Qcond =SF kS (Ts−Ta )
2πL
where SF is a shape factor, SF = ln (4z / D) , with the constraints that L ≫ D
and z>3D /2 .
The heat stored is given by
Qst =mcp
dTc
dt
(5)
where m is mass, cp is the heat capacity of the larva, and dTc / dt is the
rate of change in body temperature.
Substituting the definition of each term into Eq. (1) gives
4kL V (Tc − Ts )
dT
=SF kS (Ts−Ta )+mcp c .
R2
dt
(6)
qR2
Solving for dTc / dt and substituting Ts=Tc− 4k leads to
L
⎛ qR2
⎞ V 4k ⎛ ⎡
qR2 ⎤ ⎞ ⎤
dTc 1 ⎡
⎢SF kS ⎜
=
+Ta−Tc⎟ + 2 L ⎜Tc−⎢Tc−
⎥⎟⎥
⎝ 4kL
⎠ R ⎝ ⎣
dt mcp ⎢⎣
4kL ⎦ ⎠ ⎥⎦
(7)
2.5.3. Parameter values
All parameters are defined in Table 1. Larval volume was calculated
as V = m / ρ=πR2L . In addition, we assumed a constant larval aspect
L
L
ratio A = D = 2R of 5, which was estimated from a photograph of larval
T. dichotomus in Fig. 1b of Kojima et al. (2014). Thus, L = 2RA and
R = V /2πA .
We measured metabolic rates at room temperature (25 °C) for
larvae of a range of body masses. Measured metabolic rates fell neatly
onto the metabolic scaling lines fitted by Chown et al. (2007) to 391
insect species. Those fitted lines (ordinary least-squares and phylogenetically-corrected) had means values of 0.82 and 0.75, respectively, and
their joint confidence interval was 0.70 – 0.85. We therefore used 0.75
as the core scaling value. The effects of temperature on metabolic rate
were incorporated by fitting a sigmoid to the mean temperatureresponse curve (between 20 and 45 °C) measured in our tempera-
which simplifies to
⎤
⎛ qR2
⎞
dTc 1 ⎡
⎢SF kS ⎜
=
+Ta−Tc⎟ +Qgen ⎥ .
⎝ 4kL
⎠
dt mcp ⎣
⎦
(8)
Finally, because metabolic generation of heat itself depends on
body temperature, we make q and Qgen functions of Tc :
⎤
⎛ q (Tc ) R2
⎞
dTc 1 ⎡
⎢SF kS ⎜
+Ta−Tc⎟ +Qgen (Tc ) ⎥ .
=
⎝ 4kL
⎠
dt mcp ⎣
⎦
(9)
2.5.1. Solving Eq. (9)
Eq. (9) is a differential equation, which approaches a steady-state
body temperature dTc / dt = 0 over time if the external conditions are
79
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
Table 1
Definitions, values, and units of parameters used in the modela.
Parameter
Definition
Value (range)
Units
b
Scaling exponent of
metabolism
Mass
Metabolic rate
Aspect ratio of larva (L/D)
Radius of larva
Diameter of larva
Length of larva
Volume of larva
0.75
dimensionless
20 – 2000
variable
5
0.00876 – 0.0406
2R
0.0876 – 0.406
2.1 × 10−5 – 2.1 ×
10−3
0.05 – 0.25
950
0.5
g
W
dimensionless
m
m
m
m3
0.06 – 0.6
3421
15 − 30
W·°C−1·m−1
J·°C−1·kg−1
°C
variable
°C
variable
°C
m
M
A
R
D
L
V
z
ρ
kS
kL
cp
Ta
Ts
Tc
Depth of larva in soil
Density
Thermal conductivity of
larva
Thermal conductivity of soil
Heat capacity of larva
Temperature of the surface
of the soil
Temperature surface of the
larva
Temperature core of the
larva
m
kg·m−3
W·°C−1·m−1
Fig. 4. Mean mass and metabolic rate of T. dichotomus (black dot) plotted together with
a sample of insect metabolic rates (primarily standard metabolic rates) from 391 species
compiled and analyzed by Chown et al. (2007) and data on resting metabolic rates of the
giant burrowing cockroach Macropanesthia rhinoceros (Woodman et al., 2007; Xu et al.,
2014). Fitted lines are from Chown et al. (2007). Solid line: log10 metabolic rate
=3.26+0.82× log10 body mass; dotted line: log10 metabolic rate =3.20+0.75× log10 body
mass. Metabolic rates of M. rhinoceros are lower than for T. dichotomus in part because
the cockroaches were post-absorptive. In addition, the measurements by Xu et al. (2014)
carefully excluded periods of movement.
a
For ‘compost soil blend’ described in Ahn et al. (2009). Low values are for compost
with very low water content, and high values for high water content.
ture-ramping experiments. Values for the thermal conductivity of soil
(kS ) were based on empirically measured values for composted soil
(Ahn et al., 2009). These values differ depending on the water content
of the compost. For the core calculations, we used values from the
middle of the range.
This is most likely because of the variation and small sample size.
The average mass and metabolic rate were used to compare our
data to previous insect data on standard metabolic rates compiled by
Chown et al. (2007) and for another giant insect, the Australian
burrowing cockroach Macropanesthia rhinoceros, measured by
Woodman et al. (2007) and Xu et al. (2014). While Chown et al.
(2007) scaled all the metabolic rates to 25 °C, our data was taken at
22 °C and was not significantly altered by scaling to 25 °C, so we left
the data uncorrected. Woodman et al.’s data on M. rhinoceros were
interpolated to 25 °C using measured values at 20 and 30 °C; Xu et al.’s
data are for resting metabolic rates taken at 25 °C. It is clear (Fig. 4)
both that, along with adults of M. rhinoceros, larvae of T. dichotomus
are the largest insects measured to date and that their metabolic rates
fell close to the expected values from extrapolating standard metabolic
rates from other insects (recall that metabolic rates of T. dichotomus
include digestion, absorption, and movement). This suggests that it is
reasonable to project out metabolic rates of insects larger than actually
observed in nature for our mathematical model.
3. Results
3.1. Larval metabolic rates
The 10 larvae used in this experiment had average masses of 21.1 ±
3.5 g (mean ± S.E.M., range 15.2–26.3 g; Fig. 3). The linear regression
slope of metabolic rate on body mass was not significantly different
from zero or from one, using a linear model (confidence interval
=−0.127–1.507) and ANOVA test (P=0.087), and so it is not included.
3.2. Critical thermal maxima
Six larvae were exposed to thermal ramping to determine critical
thermal maxima. A representative recording is shown in Fig. 5. During
the temperature ramps, patterns of CO2 emission by larvae showed
several typical characteristics: a gradual increase in CO2 output
followed by one or two peaks and valleys, then followed by a final
peak, after which CO2 emissions declined rapidly and smoothly,
indicating complete loss of muscular control and death.
In most recordings, CO2 emission declined during the first 15–
20 min after the larva was introduced into the chamber, likely
reflecting that larvae were initially stressed by handling but then
became accustomed to their surroundings. After this, metabolic rate
gradually increased between 25 and 35 °C. The first peak occurred
between 35–40 °C. In some cases, a second, smaller peak appeared
before the final peak. The final peak occurred between 45 and 48 °C
(the six larvae had CTmax of 45, 45.5, 46, 47, 48, 48 °C), after which
there was a steady decline. Because larval temperature was consistently
1–1.5 °C below air temperature, larva CTmax occurred between 44–
Fig. 3. Metabolic rates of larval Trypoxylus dichotomus, estimated from rates of CO2
emission. Larval masses ranged from 15.4 to 26.3 g. In this small dataset, there was no
significant relationship between mass and metabolic rate.
80
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
Fig. 5. Typical traces showing larval activity (blue line) and CO2 emission (orange line)
during a temperature ramp, along with (lower panel) cumulative absolute difference
sums (ADS). Activity was recorded using an infrared sensor, which outputted voltage
spikes when the larva moved; because there is an arbitrary relationship between
magnitude of movement and magnitude of the spikes, the activity trace is plotted
without units. In addition, ADS curves are plotted without units because what matters is
the shape of the cumulative curve, not their actual magnitudes. Together, this information can be used to estimate CTmax. In particular, the plot shows (1; 20–35 °C) a gradual
increase in metabolic rate with temperature followed by a peak and a valley (2; 35–
40 °C). Subsequently, there is (3; 45–48 °C) a final peak ending with a (4; > 45–48 °C)
steep decline. For this larva, CTmax is estimated to be 45 °C, reflecting that CO2 emission
reaches a maximum at this temperature while activity levels fall sharply. In addition, the
emission trace becomes smooth at high temperatures, indicating that the larva has lost
control of its spiracles.
Fig. 6. Histograms of larval positions on the thermal gradient bar when (A) the ambient
window light was coming from the left and (B) from the right. Histograms were
calculated as the mean densities of each individual larva's histogram (position sampled
every 10 s for 1 or 2 h). In each panel, the grey dashed line indicates larval distributions
when there was a thermal gradient; the black line indicates when there was no thermal
gradient. Thus the x-axis indicates either temperature or position, depending on
treatment.
3.4. Analytical model of larval heat balance
47 °C.
Our model shows that, for larvae within the size range of extant
insects ( < 100 g), temperature increases due to metabolic rate are not
expected to increase larval temperature more than about 2 °C above
soil temperature (Fig. 7A, B). The main factors affecting the temperature excess above soil temperature are depth in the soil and the thermal
conductivity of the soil—greater depths and lower thermal conductivities gave higher steady-state body temperatures, all else being equal.
Low thermal conductivities are possible in dry soils (Ahn et al., 2009),
which contain a lot of insulating air spaces. Even a modest increase in
moisture, however, can raise soil thermal conductivity (kS ) to
0.2 W·°C−1·m−1 or greater. In such soils, heat is conducted more
rapidly away from its source (the larva), giving a lower steady-state
body temperature. As body size increases up to 2000 g (2 kg), larvae
begin to self-heat more significantly, again depending on a number of
factors. The highest level of self-heating ( > 6 °C) was for the largest
larvae in warm (30 °C), dry soil (low kS ). Higher soil temperatures shift
the entire larval metabolic curve up, so that relatively more self-heating
occurs in hotter soils.
3.3. Behavioral thermoregulation experiment
Preliminary experiments revealed that larvae may have been
attracted to light coming from a window in the lab. To control for
these effects, we performed experiments with the thermal gradient
oriented in both directions (hot end both toward and away from
ambient light).
Larval position on the bar depended strongly on whether or not
they were exposed to a thermal gradient. In the absence of a gradient,
larvae spent time at a relatively wide range of positions on the bar, but
clearly preferred the side toward ambient light (Fig. 6). When the light
came from the left (Fig. 6A), larvae strongly preferred the left side, as
judged by sampling 358 times (~1 h) from the mean density distribution (t-test, t=−5.29, df =357, P < 0.0001). When the light came from
the right (Fig. 6B), larvae preferred the right side (t.test, t=32.8, df
=357, P < 0.0001). By contrast, larvae exposed to the thermal gradient
strongly preferred temperatures of 28 – 30 °C (grey dotted lines in
Fig. 6A, B). In both cases, the mean spatial distributions of these larvae
differed from those of the control larvae (two-sample KolmogorovSmirnov test using 358 data points resampled from the mean density
distributions; for light from the left D =0.444, P < 0.0001; for light
from the right D =0.408, P < 0.0001). In the two experiments, the
maximum density (temperature at which the larvae spent the most
time) was 28.6 °C ± 1.87 (SEM, N =15) and 29.0 °C ± 1.27 (SEM, N
=20). Thus, larvae preferred temperatures near 29 °C.
4. Discussion
We used the large larvae of the Japanese rhinoceros beetle
(Trypoxylus dichotomus) to examine the relative roles of behavioral
thermoregulation (May, 1979; Heinrich, 1993) and metabolic selfheating in setting body temperatures. We estimated heat production
from measurements of CO2 emission and assessed the ability of larvae
81
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
There are few data in the literature available for comparison to our
results. The most relevant are several studies of brush turkey (Alectura
lathami) and mallee fowl (Leipoa ocellata) eggs (Vleck et al., 1984;
Seymour et al., 1987; Seymour and Bradford, 1992), which are
incubated in large mounds of soil and leaf litter whose decomposition
provides heat to the eggs. In a lab experiment on one brush turkey egg
and several mallee fowl eggs, Seymour et al. (1987) showed that latestage eggs had shell temperatures of 2.8 (brush turkey) or 1.9 °C
(mallee fowl) higher than surrounding soil. Compared to values
predicted by our model, these temperature increments are quite
low—especially given that the eggs are about 175 g and have metabolic
rates (at 34 – 36 °C) that are about 335 mW (335,000 µW), about 20×
higher than we measured for T. dichotomus. However, they used quite
wet soils with high thermal conductivities (0.3 – 0.5 W·°C−1·m−1)
(Seymour and Bradford, 1992) and held the eggs in small volumes of
soil in the lab such that heat was lost in all directions, both of which
would elevate rates of heat loss. In addition, they measured surface
rather than core temperatures. Thus, their data do not provide a strong
test of the model predictions we developed. Better tests would measure
core temperatures of larvae released into the wild, perhaps with small
temperature loggers implanted in their hemocoels.
For hypothetical large insect larvae, in the range of 1 – 2 kg, the
model suggests self-heating of about 1.5 – 6 °C, with the largest values
coming from the largest larvae in the hottest, driest soils. Interestingly,
larval T. dichotomus move toward CO2 signals in the soil, which has
two consequences for their distributions (Kojima et al., 2014). First,
CO2 emitted from fermenting soils (Kojima, 2015) may attract larvae to
patches of soil with locally elevated rates of respiration, which are likely
also to be warmer than surrounding patches. Second, CO2 from
conspecifics can be sensed by other larvae, causing them to aggregate
into larger groups (Kojima, 2015). Although their complicated geometry violates many of the assumptions of our simple model, such
aggregations may increase the local mass of heat-producing larval
tissues enough to make self-heating more likely.
Would self-heating—either by groups of larvae or by hypothetical
gigantic larvae—help or hurt larval performance? Although insects
living in cool environments can potentially perform better by finding
ways to warm up, this effect is unlikely to be important for underground larvae of beetles. First, because low soil temperatures depress
metabolic heat production, larvae in cool conditions are the least likely
to self-heat significantly. Second, cooler conditions often are associated
with precipitation, which would moisten soil, increase soil thermal
conductivity, and further reduce the steady-state larval temperature
excess. It is easier to construct the argument that self-heating is
harmful, from the converse of the statements above: self-heating is
greatest in the warmest soils, and it occurs more readily in dry soils,
which are likely to be associated with warmer, drier weather. Larval
CTmax is high enough, however, that self-heating is very unlikely to
push body temperatures to dangerous levels. It is more likely to push
body temperatures above the temperature preferred by larvae on the
thermal gradient bar (~29 °C). Clearly, larvae in that situation are
capable of moving to cooler areas.
Many insects are highly effective behavioral thermoregulators. A
behavioral approach to thermoregulation exploits locally available
microclimates to obtain desired body temperatures, or at least to avoid
extremes. Endothermic vertebrates obviously complement behavioral
approaches with physiological strategies that exploit metabolic heat to
attain high, stable body temperatures. Besides large flying insects, do
any other insects use the vertebrate-like strategy of thermoregulating
by controlling the production and loss of metabolic heat? We examined
this question using large beetle larvae that live in soil. Our results
suggest that their rates of heat production are too low, and rates of heat
loss too high, to achieve more than a few degrees of self-heating under
common sets of conditions. Thus, even the largest insect larvae still
must rely on behavioral thermoregulation alone.
Fig. 7. Main result of the model of larval heat balance. (A) Steady-state temperature
excess of larvae above the temperature of the soil for combinations of four soil
temperatures and two soil thermal conductivities (dark versus grey lines). Higher soil
temperatures give greater excesses because warmer larvae produce more metabolic heat.
(B) Joint effects of depth in the soil and soil thermal conductivity on the steady-state
body temperatures of a 25 g and a 250 g larva.
to choose positions on a thermal gradient bar, and we generalized our
findings using a mathematical model. Collectively, our results suggest
that large, extant larvae (20 – 30 g body mass) can self-heat by at most
2 °C, and under many common conditions (shallow depths, moister
soils) would self-heat by less than 1 °C. Larvae appear capable of
modifying their body temperatures much more significantly by behavioral thermoregulation. These results are robust because, in essence,
we stacked the deck in favor of finding an effect—by using one of the
largest larvae commercially available, which has a particularly low
surface-area-to-volume ratio for its size and which lives in decaying soil
without access to modes of heat loss that many other terrestrial insects
can use, such as convection and evaporation (Church, 1960; Prange,
1996). Moreover, active metabolic rates of larval T. dichotomus are not
particularly low for their size; they fall close to the scaling line
established from meta-analyses of mostly standard metabolic rates of
other, smaller insects (Chown et al., 2007). Thus, if any 30-g larva was
going to self-heat, T. dichotomus is a good candidate; they self-heat
modestly at best, and the model suggests this is so for larvae even
substantially larger than the ones we used.
82
Journal of Thermal Biology 62 (2016) 76–83
N.L. Cooley et al.
energetics and body temperature during free and tethered flight. J. Exp. Biol. 54,
141–152.
Heinrich, B., 1980. Mechanisms of body-temperature regulation in honey bees, Apis
mellifera. II. Regulation of thoracic temperature at high air temperatures. J. Exp.
Biol. 85, 73–87.
Heinrich, B., 1993. The Hot-Blooded Insects: Strategies and Mechanisms of
Thermoregulation. Harvard University Press, Cambridge, MA.
Hongo, Y., 2007a. Evolution of male dimorphic allometry in a population of the Japanese
horned beetle Trypoxylus dichotomus septentrionalis. Behav. Ecol. Sociobiol. 62,
245–253.
Hongo, Y., 2007b. Male–male combats of the Japanese horned beetle Trypoxylus
dichotomus septentrionalis. Nat. Insects 42, 19–22, (In Japanese).
Karino, K., Seki, N., Chiba, M., 2004. Larval nutritional environment determines adult
size in Japanese horned beetles Allomyrina dichotoma. Ecol. Res. 19 (6), 663–668.
Kingsolver, J.G., Woods, H.A., 1997. Thermal sensitivity of growth and feeding in
Manduca sexta caterpillars. Physiol. Zool. 70, 631–638. http://dx.doi.org/10.1086/
515872.
Kojima, W., 2015. Attraction to carbon dioxide from feeding resources and conspecific
neighbours in larvae of the rhinoceros beetle Trypoxylus dichotomus. PLoS One 10
(11), 1. http://dx.doi.org/10.1371/journal.pone.0141733.
Kojima, W., Ishikawa, Y., Takanashi, T., 2014. Chemically mediated group formation in
soil-dwelling larvae and pupae of the beetle Trypoxylus dichotomus.
Naturwissenschaften 101, 687–695.
Kojima, W., Takanashi, T., Ishikawa, Y., 2012. Vibratory communication in the soil:
pupal signals deter larval intrusion in a group-living beetle Trypoxylus dichotoma.
Behav. Ecol. Sociobiol. 66, 171–179.
Lighton, J., Turner, R., 2004. Thermolimit respirometry: an objective assessment of
critical thermal maxima in two sympatric desert harvester ants, Pogonomyrmex
rugosus and P. californicus. J. Exp. Biol. 207, 1903–1913. http://dx.doi.org/
10.1242/jeb.00970.
Macgregor, S.T., Miller, F.C., Psarianos, K.M., Finstein, M.S., 1981. Composting process
control based on interaction between microbial heat output and temperature. Appl
Environ. Microbiol 41, 1321–1330.
May, M.L., 1979. Insect Thermoregulation. Ann. Rev. Entomol. 24, 313–349. http://
dx.doi.org/10.1146/annurev.en.24.010179.001525.
McCullough, E.L., 2012. Using radio telemetry to assess movement patterns in a giant
rhinoceros beetle: are there differences among majors, minors, and females? J.
Insect Behav. 2012, 1–6.
Plaistow, S.J., Tsuchida, K., Tsubaki, Y., Setsuda, K., 2005. The effect of a seasonal time
constraint on development time, body size, condition, and morph determination in
the horned beetle Allomyrina dichotoma L.(Coleoptera: scarabaeidae). Ecol.
Entomol. 30, 692–699.
Prange, H.D., 1996. Evaporative cooling in insects. J. Insect Physiol. 42, 493–499.
http://dx.doi.org/10.1016/0022-1910(95)00126-3.
Rezende, E.L., Tejedo, M., Santos, M., 2011. Estimating the adaptive potential of critical
thermal limits: methodological problems and evolutionary implications. Funct. Ecol.
25, 111–121.
Robinson, W.R., Peters, R.H., Zimmermann, J., 1983. The effects of body size and
temperature on metabolic rate of organisms. Can. J. Zool. 61, 281–288. http://
dx.doi.org/10.1139/z83-037.
Schmidt-Nielsen, K., 1997. Animal Physiology: Adaptation and Environment 5th edition.
Cambridge University Press, Cambridge.
Seymour, R.S., Bradford, D.F., 1992. Temperature regulation in the incubation mounds
of the Australian brush-turkey. Condor 94, 134–150.
Seymour, R.S., Vleck, D., Vleck, C.M., Booth, D.T., 1987. Water relations of buried eggs
of mound building birds. J. Comp. Physiol. B 157, 413–422.
Soetaert, K., Petzoldt, T., Setzer, R.W., 2010. Solving differential equations in R: package
deSolve. J. Stat. Softw. 33, 1–25 . http://dx.doi.org/10.18637/jss.v033.i09, (URL)
〈http://www.jstatsoft.org/v33/i09/〉.
Terblanche, J.S., Deere, J.A., Clusella-Trullas, S., Janion, C., Chown, S.L., 2007. Critical
thermal limits depend on methodological context. Proc. R. Soc. B 274, 2935–2942.
http://dx.doi.org/10.1098/rspb.2007.0985.
Vleck, D., Vleck, C.M., Seymour, R.S., 1984. Energetics of embryonic development in the
megapode birds, mallee fowl Leipoa ocellata and brush turkey Alectura lathami.
Physiol. Zool. 57, 444–456.
Westcot, D.W., Wierenga, P.J., 1974. Transfer of heat by conduction and vapor
movement in a closed soil system. Soil Sci. Soc. Am. Proc. 38, 9–14.
Woods, H.A., Dillon, M.E., Pincebourde, S., 2015. The roles of microclimatic diversity
and of behavior in mediating the responses of ectotherms to climate change. J.
Therm. Biol. 54, 86–97. http://dx.doi.org/10.1016/j.jtherbio.2014.10.002.
Woodman, J.D., Cooper, P.D., Haritos, V.S., 2007. Cyclic gas exchange in the giant
burrowing cockroach, Macropanesthia rhinoceros: effect of oxygen tension and
temperature. J. Insect Physiol. 53, 497–504. http://dx.doi.org/10.1016/
j.jinsphys.2007.01.012.
Xu, L., Snelling, E.P., Seymour, R.S., 2014. Burrowing energetics of the giant burrowing
cockroach Macropanesthia rhinoceros: an allometric study. J. Insect Physiol. 70,
81–87. http://dx.doi.org/10.1016/j.jinsphys.2014.09.005.
Author Contributions
NLC, DJE, and HAW conceived and designed the experiments. NLC
performed the experiments, and NLC and HAW developed and
analyzed the model. NLC and HAW analyzed the data. NLC and
HAW wrote the manuscript, and DJE provided editorial advice.
Acknowledgments
This work was supported by the University of Montana and a
Davidson Honors College Undergraduate Research Award. Thanks to
Jon Harrison for critical comments on an earlier draft, and to John
Terblanche for advice on thermal ramps. A special thanks to Warren
Porter for detailed and helpful critiques of an earlier version of the
heat-balance model. We also thank two anonymous reviewers for
constructive comments on the manuscript.
Appendix A. Supporting information
Supplementary data associated with this article can be found in the
online version at doi:10.1016/j.jtherbio.2016.10.002.
References
Ahn, H.K., Sauer, T.J., Richard, T.L., Glanville, T.D., 2009. Determination of thermal
properties of composting bulking materials. Bioresour. Technol. 100, 3974–3981.
http://dx.doi.org/10.1016/j.biortech.2008.11.056.
Bakken, G.S., 1992. Measurement and application of operative and standard operative
temperatures in ecology. Am. Zool. 32, 194–216. http://dx.doi.org/10.1093/icb/
32.2.194.
Bartholomew, G.A., Casey, T.M., 1978. Oxygen consumption of moths during rest, preflight warm-up, and flight in relation to body size and wing morphology. J. Exp. Biol.
76, 11–25.
Bird, R.B., Stewart, W.E., Lightfoot, E.N., 2002. Transport Phenomena 2nd edition.
Wiley, New York.
Campbell, R., Phene, C., 1977. Tillage, Matric potential, oxygen and millet yield relations
in a layered soil. Trans. ASAE 20, 271–275. http://dx.doi.org/10.13031/
2013.35540.
Casey, T.M., 1981. Energetics and thermoregulation of Malacosoma americanum
(Lepidoptera: lasiocampidae) during hovering flight. Am. Zool. 54, 362–371. http://
dx.doi.org/10.1086/physzool.54.3.30159950.
Chown, S.L., Marais, E., Terblanche, J.S., Klok, C.J., Lighton, J.R.B., Blackburn, T.M.,
2007. Scaling of insect metabolic rate is inconsistent with the nutrient supply
network model. Funct. Ecol. 21, 282–290. http://dx.doi.org/10.1111/j.13652435.2007.01245.x.
Church, N.S., 1960. Heat loss and the body temperatures of flying insects II. Heat
conduction within the body and its loss by radiation and convection. J. Exp. Biol. 37,
186–213.
Colinet, H., Renault, D., 2012. Metabolic effects of CO2 anaesthesia in Drosophila
melanogaster. Biol. Lett. 8, 1050–1054. http://dx.doi.org/10.1098/rsbl.2012.0601.
Deutsch, C.A., Tewksbury, J.J., Huey, R.B., Sheldon, K.S., Ghalambor, C.K., Haak, D.C.,
Martin, P.R., 2008. Impacts of climate warming on terrestrial ectotherms across
latitude. Proc. Natl. Acad. Sci. USA 105, 6668–6672. http://dx.doi.org/10.1073/
pnas.0709472105.
Emlen, D.J., Warren, I.A., Johns, A., Dworkin, I., Lavine, L.C., 2012. A mechanism of
extreme growth and reliable signaling in sexually selected ornaments and weapons.
Science 337, 860–864. http://dx.doi.org/10.1126/science.1224286.
Gates, D.M., 1980. Biophysical Ecology. Springer-Verlag, New York.
Greenlee, K.J., Harrison, J.F., 2005. Respiratory changes throughout ontogeny in the
tobacco hornworm caterpillar, Manduca sexta. J. Exp. Biol. 208, 1385–1392. http://
dx.doi.org/10.1242/jeb.01521.
Greenway, H., Armstrong, W., Colmer, T., 2006. Conditions leading to high CO2 ( > 5
kPa) in waterlogged-flooded soils and possible effects on root growth and
metabolism. Ann. Bot. 98, 9–32. http://dx.doi.org/10.1093/aob/mcl076.
Harrison, J.F., Klok, C.J., Waters, J.S., 2014. Critical PO2 is size-independent in insects:
implications for the metabolic theory of ecology. Curr. Opin. Insect Sci. 4, 54–59.
http://dx.doi.org/10.1016/j.cois.2014.08.012.
Heinrich, B., 1970. Thoracic temperature stabilization by blood circulation in a freeflying moth. Science 168, 580–582. http://dx.doi.org/10.1126/
science.168.3931.580.
Heinrich, B., 1971a. Temperature regulation of sphinx moth, Manduca sexta. 1. Flight
83